advertisement
Editors Selection IGR 19-2
Progression: Combination of structural and functional measurements to predict progression
Comment by Vincent Michael Patella on:
49231 Predicting glaucomatous progression in glaucoma suspect eyes using relevance vector machine classifiers for combined structural and functional measurements, Bowd C; Lee I; Goldbaum MH et al., Investigative Ophthalmology and Visual Science, 2012; 53: 2382-2389
Bowd et al. used Relevance Vector Machine (RVM) classifiers to analyze Heidelberg confocal scanning laser ophthalmoscope (CSLO) data, Humphrey standard automated perimetry (SAP) results, and combinations of CSLO and SAP data. Their goal was to determine if progression in glaucoma suspect eyes can be predicted from clinical baselines of structural and functional measurements. RVM analyses of baseline CLSO, SAP and combinations of CLSO plus SAP data identified eyes that showed future glaucomatous progression while standard CLSO (Glaucoma Probability Score) and SAP (MD and PSD) indices did not.
Bowd and colleagues are simultaneously working on two of the big problems in glaucoma. First they are seeking to develop methods for identifying high-risk patients ‐ in this case identifying which glaucoma suspects are at elevated risk of converting to glaucoma. Second, and in pursuit of their first goal, they are developing methods for efficient combination of structural and functional data into a single metric. Clearly, their methods make available new, previously hidden diagnostic information that could be used to advance patient management. However, given the high specificities required ‐ on the order of 98 or 99% ‐ predicting conversion to glaucoma solely on the basis of baseline testing of suspects is indeed a tall order, and a task that may not yet be practical, even using this team's innovative approach. Instead, I believe that the really exciting opportunities for application of their findings lie in the construction of more effective risk assessments. Using their methods to improve our risk assessment of glaucoma suspects may turn out to be very fertile ground indeed. Generalization of these methods with the goal of predicting which glaucoma patients are most likely to progress rapidly versus slowly should also be considered. Both of these applications of the authors' methods might significantly improve the efficiency with which glaucoma suspects and glaucoma patients are managed, perhaps providing improved outcomes and more efficient use of healthcare resources.
Comments
The comment section on the IGR website is restricted to WGA#One members only. Please log-in through your WGA#One account to continue.Log-in through WGA#One
